Client pilot — in delivery, September 2026
Every repair call still depends on whoever picks up asking the right questions — at 2am or during a Monday-morning rush.
We built an AI phone agent that takes the call, works out the fault, trade and priority, and logs the job — with a staff dashboard reviewing every safety flag and repeat fault before a contractor sees it.
What the agent does
- 01
Answer
The call is picked up immediately, any time.
- 02
Diagnose
The agent establishes what's broken, where, and how urgent it is.
- 03
Log
A job record is created with trade and priority assigned.
- 04
Flag
Safety-relevant reports and repeat faults at the same property are marked for attention.
- 05
Review
Staff see every call, transcript and flagged job on a dashboard before it reaches a contractor.
The same boundary as everything we build
AI interprets. Rules and people decide what matters.
- 01
Understand
The agent listens to the tenant's own words and works out the fault, trade and priority.
- 02
Control
Safety categories, repeat-fault rules and escalation thresholds are fixed, not left to the model's judgement.
- 03
Act
A job record is created and flagged for staff review — the agent doesn't close a job or dispatch a contractor unsupervised.
Where this stands today
This is being delivered for a Scottish housing association, and the build isn't finished. We show it live on a call rather than hand out access to the dashboard, because it carries real tenant records and call transcripts.
The underlying framework is built to be reused for other housing associations' repairs lines.
Bring us your repairs line.
Book a short call and we'll walk through a real interaction, what the dashboard shows staff, and what a pilot for your organisation would involve.
No transformation programme required. Start with one process.